International Journal on Recent and Innovation Trends in Computing and Communication

Volume: 3 Issue: 4

ISSN: 2321-8169
2083 - 2089

_______________________________________________________________________________________________

Stress Detection by Measuring Heart Rate Variability

Dipali H.Patil, Garima Kumari, Pooja Daware, Vijayalaxmi Shinde, Akanksha Pran Raina
Department Of Information Technology
JSPMs Rajarshi Shahu College Of Engineering,Tathawade, Pune-33.
patil.dipali07@gmail.com, garimakumari068@gmail.com, pooja.daware14@gmail.com, vijayalaxmi.shinde9@gmail.com,
akankshapranraina@gmail.com

Abstract: In todays world one of the major leading factor to health problem is STRESS. The detection and the solution is mainly dependent on
the experience of the clinician is in detecting the factors of stress. The disadvantage of this method is that the clinicians detection may be wrong
at some stage, due to the unawareness of new problems.
The basic parameter on which stress can be identified are Galvanic Skin Response(GSR), Heart Rate(HR), Body Temperature, Blood
Pressure(BP) which provides detailed information of the state of mind of a person .These parameters vary from person to person on the basis of
certain things such as their body condition, age, gender and experience.
In our project ,we have focused on one such parameter i.e. heart rate variability(HRV) as major technique for detecting stress. HRV serves as a
substitute for vertical integration. This Vertical integration of the brain mechanism guides flexible control over behavior with peripheral
physiology and thus it provides beneficial information to understand the problems related to stress and health. In order to avoid clincians
mistake in detecting stress level, we have introduced a new hardware device which easily calculates the accurate pulse rate of a person and gives
appropriate solution to the stress level.
Keywords: Stress; Heart Rate Variability (HRV); Physiological; Galvanic skin response (GSR); Health; Vertical integration.

__________________________________________________*****_________________________________________________
I.

INTRODUCTION

Before detecting and calculating the stress of an individual,

first of all we should be well aware about the concept of
stress. Stress can be defined as A way in which a body
responds to any kind of demand. It can be caused by both
good and bad experiences. When people fell stressed by
something going around them, their bodies react by
releasing chemicals into the blood. These chemicals give
people more energy and strength, which can be a good thing
if there stress is caused by physical danger. But this can also
be a bad thing, if their stress is in response to something
emotional and there is no outlet for this extra energy and
strength. The common sources of stress are : Survival Stress,
Internal Stress, Environmental Stress and Fatigue &
Overwork.
Inspite of knowing the effect of stress on health, it is not
possible for the patient to continuously monitor their stress
levels by visiting the clinician again and again. Thus there is
a need of device that can monitor stress over regular time
intervals (weeks or months), which provides an individual
and their caretakers with data with which to monitor their
progress on their health.
In the proposed system, we have introduced a hardware
device which calculates the pulse rate to detect the stress
level. As this hardware is more beneficial since it helps to
provide the accurate result and an appropriate action to
overcome the problems related to stress.

II.

LITERATURE SURVEY

A. Definition and mechanisms of heart rate variability

The automatic nervous system (ANS) is a part of
peripheral nervous system that controls that mechanism of
our body and maintain it under stable condition. ANS
consist of two main parts: a) sympathetic nerve system
(SNS) b) parasympathetic nerve system (PNS). The SNS
prepares our body against threats so called fight or flight
response, while the PNS becomes active under
unchallenging situation, working in opposite direction and
bringing the body back to its normal state. SNS activation
increases the heart rate; on the other hand PNS activation
decreases it. By analyzing fluctuation in beat to beat periods
we can separate the role played by both branches. And this
is known as HRV analysis. HRV is a non-invasive
electrocardiographic maker which reflects on the activity of
the sympathetic and vagal components of ANS. It depicts
the total amount of variation of both HR and RR interval. In
this way, HRV analysis the tonic baseline automatic
function.
There is a continuous physiological variations of the sinus
cycles in a normal Heart with an integer ANS, which reflect
a balanced sympthovagal state and normal HRV. A
damaged heart which has suffered from myocardial necrosis,
the activity change in the afferent and efferent fibres of the
ANS and in the local nerve regulation contributes to the
resulting sympathovagal Imbalance which is reflected by a
diminished HRV.
2083

IJRITCC | April 2015, Available @ http://www.ijritcc.org

_______________________________________________________________________________________

International Journal on Recent and Innovation Trends in Computing and Communication

Volume: 3 Issue: 4

ISSN: 2321-8169
2083 - 2089

_______________________________________________________________________________________________

Fig 2. Input Output

Fig 1: Autonomous Nervous System
B. Measurement of HRV
While analyzing HRV there are a serious of measurement
which consist of successive RR interval various of sinus
variation which provides information about automatic tone.
Various factors influencing HRV include age, gender,
respiration and body position. HRV is generally performed
on the basis of 24 hour Holter recording or on a short range
in between of 0.5 to 5 min mainly focusing on
electrocardiography field.
The HRV triangular index is measured by dividing the
integral of the density distribution by the maximum density
distribution. The estimated value can be found by using a
measurement of NN intervals on a separate scale, the
measure is estimated by the value, which depends on the
length of the bin, that is, on the correctness of the separate
scale of measurement.
C. Role of neural network :
In order to understand the role of neural network, we should
first know its definition.
In information technology, a neural network is a system of
programs and data structures that approximates the
operation of the human brain. Typically a neural network is
initially trained or fed large amount of data and rules
about data relationship. A program can then tell the network
how to behave in response to an external stimulus. It
transforms inputs into outputs to the best of its ability.

The main purpose of using neural network is its ability to

learn and ability to generalize. In ability to learn NNs figure
out how to perform their function on their own and also
determine their function based only upon sample inputs.
Whereas in ability to generalize it is able to produce
reasonable outputs for inputs it has not been taught how to
deal with.
Neural Network works on 2 different layers : 1) Kohonen
Layer:Neurons in the Kohonen layer sum all of the weighted
inputs received.
Here the neuron with the largest sum outputs as 1 and the
other neurons output as 0.
2) Grossberg Layer: Each Grossberg neuron merely outputs
the weight of the connection between itself and the one
active Kohonen
Neuron.
Why Two Different Types of Layers?
Each layer has its own distinct purpose:
Kohonen layer separates inputs into separate classes
whereas inputs in the same class will turn on the same
Kohonen neuron.
Grossberg layer adjusts weights to obtain acceptable
outputs for each class.
The technology especially artificial neural network (ANN)
techniques could result in reducing cost, time, medical error
and need of human expertise. Neural network model is used
to detect stress with the help of physical symptoms and risk
factors. Once the neural network model is trained, it will
predict the possibility of stress.
Artificial Neural Networks are non-linear mapping
structures based on the function of the human brain.
Artificial Neural Networks can identify and learn correlated
patterns between input data sets and corresponding target
values.
2084

IJRITCC | April 2015, Available @ http://www.ijritcc.org

_______________________________________________________________________________________

International Journal on Recent and Innovation Trends in Computing and Communication

Volume: 3 Issue: 4

ISSN: 2321-8169
2083 - 2089

_______________________________________________________________________________________________
E. Geometric methods

Fig 3. Architecture for diagnosis of stress

D. Time domain analysis
In Time Domain Analysis, it helps to measure the changes
in heart rate over a time period. In ECG recording in which
is of 24 hours, normal RR intervals and QRS complex is
detected, because of sinsus depolarization or the
instantaneous heart rate are then determined. The statistical
time domain indices are divided in to two categories i.e.
beat- to-beat intervals and intervals derived from the
differences between adjacent NN intervals.
Table 1 summarizes two different categories, the first
category consists of parameters like SDNN, SDANM and
SD and these of second category are RMSSD and PNN50.
SDNN is a global index of HRV, which reflect the long
term components responsible for variability in the recording
period. SDANN is an index of variability having an average
of 5 min interval over 24 hours. SD is consider to reflect the
day night changes of HRV .RMSSD and PNN50 are
common parameters which are based on the differences in
the interval. These are not dependent on day and night
variations and respond to short term HRV changes. As
RMSSD is more stable hence it should be preferred for
clinical use.

The conversion of sequences of NN intervals when derived

and constructed gives geometric methods. There are
different geometrical forms available which allow access to
HRV: the 24-hour histogram, the HRV triangular index its
modification, the triangular interpolation of NN interval
histogram and the method based on Lorendz or Poincare
plots. The relationship between the total number of RR
interval detected and the 24-hour RR interval variation is
accessed by 24-hour histogram. The triangular HRV index
has the measure peak of histogram as a triangular with its
height corresponding to the most frequently observed
duration of RR intervals, its baseline width corresponding
the amount of RR interval variability and its area
corresponding to the total number of all RR interval used to
constructed. By the quality of the recorded data, geometrical
methods are less affected and they may provide on alternate
to less easily obtainable statistical parameters.
F. Frequency domain analysis
The periodic oscillations of heart rate signal decompose at
different frequencies and amplitudes are described by
frequency domain analysis, and it provide information on
the amount of their relative intensity in the hearts sinus
rhythm. Power spectral analysis is obtained when white light
passes through a prism which results in different light of
different colour and wavelength. The analysis of power
spectral can be done in two ways: 1) Using non parametric
method, the fast fourier transformation (FFT), which is
distinguished by discrete Peaks for the several frequency
components. 2) Using parametric method, the auto
regressive model estimation, which results in a continuous
smooth spectrum of activity. The parametric method more
complex and needs verification of the suitability of the
model choosen whereas FFT is simple and rapid methods.

2085
IJRITCC | April 2015, Available @ http://www.ijritcc.org

_______________________________________________________________________________________

International Journal on Recent and Innovation Trends in Computing and Communication

Volume: 3 Issue: 4

ISSN: 2321-8169
2083 - 2089

_______________________________________________________________________________________________

2086
IJRITCC | April 2015, Available @ http://www.ijritcc.org

_______________________________________________________________________________________

International Journal on Recent and Innovation Trends in Computing and Communication

Volume: 3 Issue: 4

ISSN: 2321-8169
2083 - 2089

_______________________________________________________________________________________________
III.

ARCHITECTURE

3.

4.

Fig 4: Architecture
IV. MAIN MODULES

5.

6.
7.
Fig 5: Block Diagram
The modules in our project includes:
1.
2.

Pulse Sensor : This is the input device which takes

the pulses as the input from the user.
Hardware : The hardware is a combination of
several small small components. These components
include capacitor, diodes, LED, LED display
screen, USB cable and microcontroller LPC2148.
The LPC2148 microcontrollers is based on a 16bit/32-bit ARM7TDMI-S CPU with real-time
emulation and embedded trace support, that
combine microcontroller with embedded highspeed flash memory ranging from 32 kB to 512 kB.
Due to their tiny size and low power consumption,
LPC2148 are ideal for applications where
miniaturization is a key requirement, such as access

control and point-of-sale. Serial communications

interfaces ranging from a USB 2.0 Full-speed
device, multiple UARTs, SPI, SSP to I2C-bus and
on-chip SRAM of 8 kB up to 40 kB, make these
devices very well suited for communication
gateways and protocol converters, soft modems,
voice recognition and low end imaging, providing
both large buffer size and high processing power.
System : It is our computer system which displays
the result. In our project we have separate login for
both doctor and patient. Where first the patient
registers by filling their personal details and with
the help of hardware they can easily determine their
pulse rate as well as stress level. Now if the
particular person is below normal or highly tensed
they need to refer the doctor and take the
appropriate solutions.
Process : It is a method where the artificial
intelligence performs two actions : Training and
Testing.
There are two main types of training:
a) Supervised Training: It Supplies the neural
network with inputs and the desired outputs
and it responses of the network to the inputs is
measured. The weights are modified to reduce
the difference between the actual and desired
outputs.
b) Unsupervised Training : It only supplies
inputs. The neural network adjusts its own
weights so that similar inputs cause similar
outputs. And the network identifies the
patterns and differences in the inputs without
any external assistance.
Artificial Intelligence: It performs training and
testing, where training is done by using data
stored in the dictionary and testing is done by
various input.
Dictionary: It contains all the pulse ranges with
their corresponding stress level.
Decision: It helps to decide the stress level
based on the users pulse rate by referring the
details provided by the process.

8.

Ation: Depending on the stress level of the

user appropriate actions are taken to cope up
with their stress.
IV.

LIMITATION OF STANDARD HRV

MEASUREMENTS

As HRV measurement deals with RR interval variations, its

measurement is limited to only those patient with low
number of ectopic beats and patients in sinus rhythm. Due to
this almost 20-30% of high risk post-MI patients are
neglected from any HRV analysis due to episodes of atrial
arrhythmias, particularly atrial fibrillation.

2087
IJRITCC | April 2015, Available @ http://www.ijritcc.org

_______________________________________________________________________________________

International Journal on Recent and Innovation Trends in Computing and Communication

Volume: 3 Issue: 4

ISSN: 2321-8169
2083 - 2089

_______________________________________________________________________________________________
pertaining to HRV measured after MI, suggesting that this
technique may be not enough by itself to adequately risk
stratify these high risk patients.
The combination of HRV with other risk stratifies, which
includes LVEF, NSVT, LP and BRS may increase the
overall analytical accuracy. In recent times an approach
using a variety of noninvasive and invasive tests

V.

HRV AND EMOTIONAL REGULATION

The ability to control emotion is related to the ability to

shape affective brain pressure flexibly in response to
changing context. Emotions represent on individuals vision
of personally relevant environment interactions, which
includes not only the challenges and threats but also the
ability to respond to them. Emotion reflect the status of an
individuals ongoing adjustment to repeatedly changing
environment demands. The role of HRV in emotional
regulation is seen at two different levels. The first level is at
tonic level where individual differences in resting HRV are
associated with the differences in their emotional regulation.
People with high level of resting HRV produce context
appropriate emotional responses as indexed by fear
potentiated startle responses, phasic heart rate responses,
self reported emotional responses, emotions modulated
startle responses, as compared to people with low level of
resting HRV. The second level is the phasic level where the
HRV values increases where there is a successful regulation
of emotion during emotional task.
VI.

CURRENT LIMITATION AND FUTURE

DIRECTIONS FOR THE USE OF HRV

A large number of experimental and clinical studies which

are earlier published for the measurement of HRV is still a
research technique and still has not proved to be a routine
clinical tool. There are many reasons behind this, Firstly, the
physiopathological mechanism of HRV which establishes a
direct link between mortality and reduce HRV which is still
not fully clear. Secondly, the application of HRV assessment
in clinics is limited due to lack of standard methodology
which is due to the variability of the parameters varying
with respect to age, gender, drug interferences and
associated diseases. Third, even with relative evidences of
the vigorous character of parameters, there is still no
consensus about the most accurate parameter. Fourth, the
specificity, sensitivity and positive analytical accuracy of
HRV are limited. Mostly, its positive analytical accuracy is
modest, which ranges from 14 to 40%.While it is also
having negative analytical value which ranges from 77 to
98%. At last contradictory results have been found

in a stepwise fashion was proposed. In stage one LP and

LVEF were obtained whereas in stage two with the help of
an ambulatory 24-hour ECG recording for the
documentation of complex ventricular arrhythmias and for
the measurement of HRV, and in stage three use of an
electrophysiology study with potential induction of
ventricular tachycardia.
VII.

RESULTS AND DISCUSSIONS

Fig 6: Login

Fig 7: Share Doctor

2088
IJRITCC | April 2015, Available @ http://www.ijritcc.org

_______________________________________________________________________________________

International Journal on Recent and Innovation Trends in Computing and Communication

Volume: 3 Issue: 4

ISSN: 2321-8169
2083 - 2089

_______________________________________________________________________________________________
analysis. We have also focused on the relation between
HRV and emotional regulation. It is our hope that this
review highlights the different analysis for HRV
calculations. As we are using digital pulse sensor, it could be
less time consuming and more accurate than the older
techniques. Our expectation is to maintain accurate stress
using Artificial Neural Network (ANN), we are going to
analyze the records provided by the hardware machine and
depending on that reading we are going to determine stress
level.

XI. REFERENCES

Fig 8: Patient Information

[1] Stress Management Using Artificial Intelligence V.J.

Madhuri , Madhumitha R.Mohan, Kaavya.R.
[2] Heart
rate
variability:
a
noninvasive
electrocardiographic method to measure the autonomic
nervous system Juan Sztajzel September 2004.
[3] A meta-analysis of heart rate variability and neuro
imaging studies: Implications for heart rate variability as
a marker of stress and health Julian F. Thayera,b,_,
Fredrik hsc, Mats Fredriksonc, John J. Sollers III d, Tor
D. Wagere.
[4] Using Heart Rate Monitors to Detect Mental Stress
Jongyoon Choi and Ricardo Gutierrez-Osuna, 2009 Body
Sensor Networks.
[5] Google images .

Fig 9: Result
VIII.

FUTURE SCOPE

This paper is must read for everyone but specially the one
who is suffering from hypertension so that they can get
different solutions to overcome their daily stress and can
live their life happily without stress. Here are some
prevention strategies to control stress.
1.They need to increase their body awareness and relaxation
through biofeedback, meditation, or yoga.
2. Learn to manage anger .
3. Communicate effectively with family members, peers,
supervisors, and citizens.
SUMMARY AND CONCLUSION
In this paper, we have presented different analysis for Heart
Rate Variability (HRV).We have shown the correlation
between two analysis. This review supports for time-domain
2089
IJRITCC | April 2015, Available @ http://www.ijritcc.org

_______________________________________________________________________________________